• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.04a
• /
• pp.65-68
• /
• 2004

복잡한 문제 학습을 위해 여러 가지 형태의 모듈라 네트워크의 구조가 제시되어 왔다. 그 중 엑스퍼트 네트워크와 게이팅 네트워크로 구성된 Mixtures of Experts network은 복잡한 문제를 단순한 문제들로 분해하고, 각각의 엑스퍼트 네트워크가 분해된 단순한 문제를 학습하여 결과를 도출함으로써, 국소적 지역해의 위험을 방지하고 보다 정확한 학습을 가능하게 한다. 그러나 엑스퍼트 네트워크의 수렴은 게이팅 네트워크의 수렴에 많은 영향을 받게 되고, 모든 복잡한 데이터에 대한 엑스퍼트 네트워크의 기여도를 학습하는 게이팅 네트워크는 역전파 알고리즘에 의한 학습 방법으로는 수렴 속도가 떨어진다. 본 논문에서는 게이팅 네트워크를 칼만필터로 학습하여 복잡한 문제에 대한 강건성은 유지하고 보다 빠른 수렴이 가능한 방법을 제시하고자한다.

• KSCI Review
• /
• v.2 no.2
• /
• pp.41-50
• /
• 1996

• Journal of Korean Electronics
• /
• v.5 no.1
• /
• pp.70-73
• /
• 1985

• 프린팅코리아
• /
• s.57
• /
• pp.144-145
• /
• 2007

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.11 no.6
• /
• pp.1014-1025
• /
• 1987

This paper proposes an approach of building an expert system for designing gear drives. In terms of gear drives the relationship between expert systems and data base systems is explained and a data base system for design, failure detection, and trouble-shooting of gear drives is built on a personal computer by the data manipulation language of dBASE III. This paper proposes a method how to build a data base system and what to consider to build a data base system for gear drives and two kinds of data base systems of design and failure detection of gear drives and trouble-shooting for noise and vibration of gear drives are built. The two data base systems are very effective for the retrieval of information for the gear design, and they will be considerably helpful for the gear design.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1993.04b
• /
• pp.289-293
• /
• 1993

연소가공은 가공에 관계하는 변수가 많기 때문에 트라블 요인을 쉽게 파악하기 힘들다. 트라블의 원인을 탐색해서 처리하는 경우에도 정량적으로 정확한 값을 User에게 제공하기는 어렵다. 따라서 대부분 생산시스템에 활용할때 숙련자가 행하는 것 같이하기 위한 유용한 수단이 엑스퍼트 시스템( Expert System)이다. 본 연구는 연삭가공의 Burning과 Chatter vibration의 진단 . 처리에 대하여 폴트 진단모델 과 계층분석법을 이용하여 Prototype적인 시스템을 구축하였다. 또한, 각 노드에 대한 데이타 분석에 있어, 계층분석법을 이용함으로써 정량적이고 애매한 지식을 객관적으로 처리 하였으며, 각각의 노드에 대하여 중요도를 부여함으로써 데이타 베이스와 추론시간의 성력화를 도모하였다.

• 전기의세계
• /
• v.38 no.11
• /
• pp.48-54
• /
• 1989

제철소에 대한 설비진단은 발전을 거듭하면서 금세기에 들어서는 AI를 이용한 시스템 구축이 이루어지고 있다. 그중에서도 고로에 대한 엑스퍼트 시스템은 상당한 실효를 거두고 있다. 그 이유는 고로는 제철소의 최상류에 위치하며 제품 공정에의 선철의 안정 공급이 큰 역할로 되어 있기 때문에 고로에 대한 많은 연구와 노력이 집대성된 까닭이다. 하여튼 제철소 내에는 어느 공장하나 중요치 않은 것이 없다. 그중에서도 이번에 소개하고자 하는 것은 제철소 제강공장에 대한 설비 진단 시스템이다. 제강공장은 직접적인 생산품을 만들어 내는 곳은 아니지만 제품의 품질을 좌우하는 중요도가 높은 공장이다.

• Water for future
• /
• v.23 no.4
• /
• pp.477-486
• /
• 1990

The domain of contaminated groundwater flow is a broad and multidisciplinary field requiring expertise in engineering, geology, chmistry and toxicology, and is an ideal area for the application of Expert System. The Expert System which is developed in this research can assist user to find possible remedial actions in case that the groundwater was contaminated with toxic pollutants. Documentation including the degree of toxicity, the possibility of chemical reaction and concentration of pollutant can be supported also. Prolog, and artificial intelegence programming language, is used to implement the prototype expert system. This expert system can explicitly advise users about contaminants' toxicity, possiblity reaction with other chemicals and their concentrations.

• Journal of Bio-Environment Control
• /
• v.30 no.1
• /
• pp.27-36
• /
• 2021

A capacitive water sensor was developed to measure the capacitance over a wide part of a substrate using an insulated electrode plate (30 cm × 10 cm) with copper and Teflon attached on either side of the substrate. This study aimed to convert the capacitance output obtained from the condenser-type capacitance sensor into the substrate water content. The quantification experiment was performed by measuring the changes in substrate water weight and capacitance while providing a nutrient solution and by subsequently comparing these values. The substrate water weight and capacitance were measured every 20 to 30 seconds using the sensor and load cell with a software developed specifically for this study. Using a curve-fitting program, the substrate water content was estimated from the output of the capacitance using the water weight and capacitance of the substrate as variables. When the amount of water supplied was increased, the capacitance tended to increase. Coefficient of variation (CV) in capacitance according to the water weight in substrate was greater with the 1.0 kg of water weight, compared with other weights. Thus, the fitting was performed with higher than 1.0 kg, from 1.7 to 6.0 kg of water weight. The correlation coefficient between the capacitance and water weight in substrate was 0.9696. The calibration equation estimated water content from the capacitance, and it was compared with the substrate water weight measured by the load cell.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.21 no.12
• /
• pp.52-59
• /
• 2020

Currently, the domestic electricity rates for houses are charged by applying a progressive level according to monthly electricity usage. Electricity rates rise sharply wWhen the amount of electricity used is large, electricity rates rise sharply. The standardized electricity rate progressive system has limitations in that it lacks consideration of the consumers' power usage patterns and limits consumers' their options. Accordingly, the Ministry of Trade, Industry and Energy and the Korea Electric Power Corporation have been demonstrating the basis of a rate system for housing, which is a method of charging electricity according to the amount of electricity used by season and time. In this paper, 10 electricity usage patterns were derived through from AMI data analysis for 5 five years of 362 apartment complexes located in metropolitan cities. The patterns were, and then applied to the existing domestic electricity rate and time-by-time rates applied to demonstrations, and by time-by-time rates in the US and Australia. The effect of the optional rate by pattern was compared and analyzed. As a result, it was confirmed that benefits occurred in five5 patterns compared to existing rate plans, and the electricity rates increased in 5 five patterns, and t. This phenomenon shows the same phenomenon withis the same as the overseas rates, including domestic rates being demonstrated.